Figure 1.
Small molecule evolution by DNA-programmed combinatorial chemistry.
(A) A degenerate library of DNA sequences is chemically translated into small molecule-DNA conjugates. The attached small molecule corresponds to the structure encoded by the DNA “gene”. The translated molecules are selected for a desired trait, such as binding to a protein of interest. The encoding DNA is amplified and diversified. The cycle is iterated to yield small molecules with the selected property. (B) Translation of a single coding position by DNA-programmed combinatorial chemistry. A degenerate DNA library is split by hybridization to oligonucleotide-conjugated resins arrayed in a 384-well cassette (the anticodon array). The DNA is then transferred in a one-to-one fashion onto an anion-exchange chemistry array using a mesofluidic Southern blotter. The transferred DNA is subjected to a chemical coupling step, with a different chemical building block used in each well. The DNA is eluted and pooled. These operations are repeated until all of the coding positions in the gene have been read.
Figure 2.
Assembly and characterization of a degenerate DNA library.
(A) Structure of the DNA genes. The DNA genes consist of five variable positions (VA, VB, VC, VD, VE) flanked by six constant regions (ZA, ZB, ZC, ZD, ZE, ZF). The gene library was assembled with 384 different codon sequences at each of the VA–VD positions and 10 possible codon sequences at the final VE position. (B) Test of hybridization specificity. A sub-library of DNA genes with 32 distinct codon sequences at the B coding position (corresponding to the third and fourth columns of the B anticodon array) and 384 distinct codon sequences at the other coding positions (1194 possible codons) was hybridized to the B anticodon array, which holds 384 different oligonucleotide-conjugated resins. After hybridization, the array was imaged using a phosphor screen. Strong and roughly equal signal intensities are observed at the wells in the third and fourth columns of the anticodon array. No signal over background noise is observed at the other wells of the array.
Figure 3.
Accurate routing of DNA using mesofluidic devices.
(1) Using the mesofluidic pump, four radiolabeled DNA sequences (ZA–A1, ZB–B2, ZC–C10, and ZD–D7) were hybridized to an anticodon array filled with 96 oligonucleotide-conjugated resins. The four spots at well positions A1, E19, G13, and O3 of the array (indicated by red boxes) were filled with resins A1′, B2′, C10′ and D7′ respectively. (2) Hybridized DNA was transferred to an anion-exchange chemistry array using a mesofluidic Southern blotter. (3) The DNA was eluted from the anion-exchange chemistry array, pooled, and split again using an anti-codon array containing the same 96 resins in a different order: well positions G9, G15, I9, and I15 of the array (indicated by red boxes) were filled respectively with the A1′, B2′, C10′, and D7′ resins. The arrays were imaged at each step.
Figure 4.
A single DNA gene comprising the A1, B1, C37, D1 and E1 codons was hybridized to the A anticodon array, transferred to an anion-exchange chemistry array, eluted from that array, and then hybridized to the B anticodon array. Nucleic acid was isolated from the upper left well (corresponding to the B1′ anticodon) and quantified. Approximately 85% of the routed DNA was recovered after this routing cycle.
Figure 5.
Proof-of-principle chemical selection.
A degenerate library of DNA genes lacking one of the 1546 codons (C37) was mixed at 383 parts to 1 part with a short gene encoding C37 and missing one coding (VD) and one constant (ZD) position. The DNA solution was split over the C anticodon array and then blotted onto an anion-exchange chemistry array. The amine “synthetic nucleus” on the DNA in each well was acylated with chloroacetic acid. Propylamine was coupled to the DNA molecules in 383 of the wells, while the DNA in well G1 (corresponsing to the C37′ anticodon) was coupled to biotin hydrazide. The resulting peptoid-DNA conjugates were eluted, pooled and selected for binding to streptavidin-coated magnetic beads. The isolated material was amplified and analyzed on a 3% agarose gel, revealing a 13,000-fold enrichment of the sequence encoding the biotin monomer.
Figure 6.
Analysis of a routed chemical translation.
A single DNA gene comprising the A1, B1, C37, D1 and E1 codons was routed to the upper left well of the A anticodon array (hybridization of the first coding position to the A1′ anticodon) where a propylamine peptoid monomer was added. Similar reads of the second coding position (hybridization to the B1′ anticodon and addition of cyclopropylamine) and third coding position (hybridization to the C37′ anticodon and addition of benzylamine) completed the DNA-programmed synthesis. For comparison purposes, an identical synthesis was carried out without routing. Synthetic intermediates and products were digested with phosphodiesterase I and analyzed by reverse-phase HPLC. The major peaks for the routed products were isolated and analyzed by LC-MS (product 2 [M+H]+, expected 868.40, observed 868.75; product 3 [M+H]+, expected 965.45, observed 965.86; product 4 [M+H]+, expected 1112.52, observed 1112.97).